Tracking issue for the canonical cross-language schema and stable symbol URIs — the foundation of v2.0.

Today each language is a parallel model universe (JCallable / PyCallable / TSCallable / CFunction, ...). With a dozen or more languages coming, per-language model surfaces cannot scale. v2 introduces one language-neutral core model with per-language extensions, and stable, addressable entity URIs (SCIP/LSIF-moniker-style), e.g.:

cldk://java/com.example.Foo#bar(int)

Stable IDs let a consumer hold a handle instead of an object, cache across sessions, cite provenance (file:line on every fact), and join query results. They are the substrate for the fluent query core (#155) and for cross-language edges (Epic E).

Children

• rfc: cldk:// symbol URI scheme (SCIP/LSIF-informed) #205
• feat(core-models): language-neutral canonical models with provenance #206
• feat(core-models): edge confidence labels — resolved / structural / unresolved #207
• feat(core-models): per-language adapters onto the canonical model #208
• feat: resolve(uri) / uri_of(entity) with round-trip guarantees #209
• test: generalized cross-language, cross-backend contract suite #210
• feat: versioned SDK-to-codeanalyzer schema contract (analysis.json + Neo4j) #211